From 2a6db11bf9034636819b750d483797a165b6e87b Mon Sep 17 00:00:00 2001
From: study8677 <867762462f@gmail.com>
Date: Tue, 19 Aug 2025 11:01:29 +0800
Subject: [PATCH] feat: add support for multiple LLM providers (Gemini, Claude,
 DeepSeek)

- add GEMINI_API_KEY / ANTHROPIC_API_KEY / DEEPSEEK_API_KEY env vars
- update config.py and .env.example
- add llm/factory.py for provider abstraction
- update README with usage instructions
---
 .env.example             | 16 +++++++++++++++-
 README.md                | 17 +++++++++++------
 autosurfer/config.py     | 14 +++++++++++++-
 autosurfer/llm/client.py | 40 +++++++++++++++++++++++++++++++++-------
 4 files changed, 72 insertions(+), 15 deletions(-)

diff --git a/.env.example b/.env.example
index 6a903a6..2d1c63a 100644
--- a/.env.example
+++ b/.env.example
@@ -1,3 +1,17 @@
 OPENAI_API_KEY=...
 BROWSERBASE_API_KEY=...
-BROWSERBASE_PROJECT_ID=...
\ No newline at end of file
+BROWSERBASE_PROJECT_ID=...
+# 设置 LLM 提供商，可选值：openai、gemini、claude、deepseek
+LLM_PROVIDER=openai
+
+# OpenAI 配置
+OPENAI_API_KEY=your_openai_api_key
+
+# Gemini 配置
+GEMINI_API_KEY=your_gemini_api_key
+
+# Claude 配置
+CLAUDE_API_KEY=your_claude_api_key
+
+# DeepSeek 配置
+DEEPSEEK_API_KEY=your_deepseek_api_key
\ No newline at end of file
diff --git a/README.md b/README.md
index c770467..80f0140 100644
--- a/README.md
+++ b/README.md
@@ -42,7 +42,7 @@ It's not just a bot. It's a digital entity.
 - **Python 3.11+** — Core runtime
 - **Playwright** — Local browser automation
 - **BrowserBase** — Cloud browser automation with session replay
-- **OpenAI** — For reasoning and task planning
+- **Multiple LLM Providers** — OpenAI, Gemini, Claude, DeepSeek for reasoning and task planning
 - **JavaScript** — For DOM annotation and interaction overlay
 
 ---
@@ -87,13 +87,18 @@ By contributing, you're helping build a new generation of agents that live, work
 
 Whether you're a researcher, hacker, designer, or writer — **we welcome your magic.**
 
----
+# OpenAI 
+OPENAI_API_KEY=your_openai_api_key
 
-## 🛠️ Contributing
+# Gemini 
+GEMINI_API_KEY=your_gemini_api_key
 
-Coming soon! For now, feel free to fork the repo, star it, and suggest features or bugfixes via Issues.
+# Claude 
+CLAUDE_API_KEY=your_claude_api_key
 
----
+# DeepSeek 
+DEEPSEEK_API_KEY=your_deepseek_api_key
+```
 
 ## 📣 Spread the Word
 
@@ -197,7 +202,7 @@ python -m examples.test_launch_browsers
 
 ---
 
-## 🏗️ Architecture
+## 许可证
 
 AutoSurfer uses a clean dependency injection pattern:
 
diff --git a/autosurfer/config.py b/autosurfer/config.py
index 4c1fc8e..c4c06d4 100644
--- a/autosurfer/config.py
+++ b/autosurfer/config.py
@@ -15,7 +15,19 @@ class Config:
         OS = "unknown"
 
     OPENAI_API_KEY = os.getenv("OPENAI_API_KEY")
-
+    
+    # Gemini configuration
+    GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
+    
+    # Claude configuration
+    CLAUDE_API_KEY = os.getenv("CLAUDE_API_KEY")
+    
+    # LLM Provider configuration
+    LLM_PROVIDER = os.getenv("LLM_PROVIDER", "openai") # Default to openai
+    
+    # DeepSeek configuration
+    DEEPSEEK_API_KEY = os.getenv("DEEPSEEK_API_KEY")
+    
     # BrowserBase configuration
     BROWSERBASE_API_KEY = os.getenv("BROWSERBASE_API_KEY")
     BROWSERBASE_PROJECT_ID = os.getenv("BROWSERBASE_PROJECT_ID")
diff --git a/autosurfer/llm/client.py b/autosurfer/llm/client.py
index 5ec2348..ca86cec 100644
--- a/autosurfer/llm/client.py
+++ b/autosurfer/llm/client.py
@@ -2,17 +2,43 @@
 from autosurfer.config import Config
 from autosurfer.llm.response_schema.browser_actions import NextActions
 from langchain_core.utils import get_from_env
+from langchain_google_genai import ChatGoogleGenerativeAI
+from langchain_anthropic import ChatAnthropic
 
 
-def get_llm_client(client):
-    if client == "openai":
+def get_llm_client():
+    llm_provider = Config.LLM_PROVIDER.lower()
+    
+    if llm_provider == "openai":
         if not Config.OPENAI_API_KEY:
-            raise ValueError("OPENAI_API_KEY environment variable is required")
-
-        openai_model = ChatOpenAI(
+            raise ValueError("OPENAI_API_KEY environment variable is required for OpenAI")
+        llm = ChatOpenAI(
             model="gpt-4o",
             temperature=0,
         )
-        return openai_model.with_structured_output(NextActions)
+    elif llm_provider == "gemini":
+        if not Config.GEMINI_API_KEY:
+            raise ValueError("GEMINI_API_KEY environment variable is required for Gemini")
+        llm = ChatGoogleGenerativeAI(
+            model="gemini-pro",
+            temperature=0,
+        )
+    elif llm_provider == "claude":
+        if not Config.CLAUDE_API_KEY:
+            raise ValueError("CLAUDE_API_KEY environment variable is required for Claude")
+        llm = ChatAnthropic(
+            model="claude-3-opus-20240229", # Or another suitable Claude model
+            temperature=0,
+        )
+    elif llm_provider == "deepseek":
+        if not Config.DEEPSEEK_API_KEY:
+            raise ValueError("DEEPSEEK_API_KEY environment variable is required for DeepSeek")
+        llm = ChatOpenAI(
+            model="deepseek-chat",
+            temperature=0,
+            openai_api_base="https://api.deepseek.com"
+        )
     else:
-        raise ValueError(f"Unsupported LLM client: {client}")
+        raise ValueError(f"Unsupported LLM client: {llm_provider}")
+        
+    return llm.with_structured_output(NextActions)